Fingerprint Indoor Positioning Method with Deep Learning

A variety of application services have been developed one after another with information and communications technologies coming to the fore over the past years. In this regard, the application services of indoor positioning, for example, indoor positioning navigation, ad push and data analysis, for users mostly staying in indoor environments per day are diversified and promotes economic efficiency significantly. However, indoor positioning is still not accurate enough because an indoor environment is rich in all kinds of noises and more complicated than an outdoor environment. In this research, indoor positioning is embodied through Beacons based on Bluetooth 4.0 and particularly fingerprint indoor positioning with deep learning incorporated is adopted for promotion of accuracy of indoor positioning. Differing from ordinary fingerprint positioning during which a fingerprint map is trained through deep learning in the offline stage only, fingerprint indoor positioning in this research is characterized that signals received in the positioning stage are recognized through deep learning for better validity of signals received and further checked with a trained fingerprint map for estimation of a current location.